As a control core of an entire communications network, a core network needs to connect to various peripheral networks and devices. All network services need to be controlled and processed or forwarded by a core network element, and a network service provided by the core network element is irreplaceable. Therefore, when the core network element needs to be upgraded, a system that is processing a network service needs to be upgraded while normal processing of the network service needs to be ensured. That is, hitless upgrade of the core network needs to be implemented, and the service is not interrupted during upgrade of the system or only a brief interruption (for example, within 10 s) is allowed.
In the conventional art, hitless upgrade is usually performed on a core network element by using the following method: Processes in the network element are classified into a process on an active plane and a process on a standby plane; during upgrade, the standby plane is first upgraded to a new version (no service is processed on the standby plane); after the standby plane is upgraded, status switching is performed on the standby plane and the original active plane, and all services are switched from the original active plane to the original standby plane (a new active plane); then, the original active plane is upgraded. In the upgrade process, a very short service interruption (theoretically, a service interruption time is approximately 10 s) is generated only when the services are switched from the active plane to the standby plane that is upgraded to the new version, so as to implement hitless upgrade.
However, in a process of implementing the embodiments of the present invention, it is found that the foregoing hitless upgrade method has the following problems: Device configurations and networking of an existing network vary widely, an actual scenario in the existing network cannot be simulated at a research and development phase, and many problems of a network element of a new version can be found only during upgrade. Consequently, during the upgrade, after the network element switches all services to the new version, an overall situation may be affected once a problem occurs during the upgrade to the new version.